Abstract: To effectively prevent the impact of agrometeorological disasters on the production of apricot in Tianshui, this study used the statistical method and risk principle to extract the main disaster factors and constructed the model of meteorological disaster risk assessment. The analysis showed that the main disaster factors that heavily affected the apricot production were: drought in early March before flowering, freezing in early and middle April, rainy and sunlight shortage in late July during fruit stage, less rainfall and drought in May. Compared with that of 1982—1991, the 10 year average of climate yield dynamic relative deviation decreased by 29.9% in 1992—2001 and decreased by 7.8% in 2002—2011. The meteorological disaster assessment showed that the disaster years were 16 years with an accuracy rate of 94%. The accuracy rate of mild, moderate, severe and the worst disaster was 50%, 88%, 100%, 67%, respectively. The accuracy rate of mild and the worst disaster were lower but the evaluation level varied only one level comparing with the real level. The disaster evaluation accuracy was high and the result was ideal.